UK FASHIoN: feasibility study of a randomised controlled trial of arthroscopic surgery for hip impingement compared with best conservative care

Introduction

The prevalence in the general population of symptomatic FAI (FAI hip shape morphology and concurrent hip symptoms) is not known. Similarly the proportion of these patients who require surgery is not known. The published literature has an overwhelming focus on the surgical management of FAI. 1,31,32 Obtaining an understanding of the amount of surgery that is being undertaken for FAI would help to understand the burden of symptomatic FAI and help to determine the likely pool of eligible patients for a RCT.

Objective

To estimate the frequency and types of FAI surgery being undertaken in the UK within the NHS.

Materials and methods

A list of all NHS hospital health boards and trusts within England, Northern Ireland, Scotland and Wales was compiled using the NHS online resource: www.nhs.uk. All hospital trusts and health boards were contacted by telephone to determine if they had an orthopaedic service and within each orthopaedic service the number of orthopaedic departments that made up that service. Clinical directors for these departments/services were then contacted by letter requesting the names and contact details of all FAI surgeons within their department/service. Each FAI surgeon identified received a letter requesting the following workload data for the financial year 2011/12:

• hip arthroscopies performed within the NHS over the last 12 months

• number of hip arthroscopies performed for FAI within the NHS over the last 12 months

• number of open surgical procedures performed for FAI within the NHS over the last 12 months.

Clinical directors and surgeons were contacted repeatedly by letter, e-mail and telephone and through their secretaries. When a consultant’s practice spanned less than 12 months, the results were not rescaled; instead, conservative estimates were obtained for the full 12 months by keeping case numbers the same, no matter the period of practice. When consultants provided a range, this was recorded and final calculations were made based on the lowest figure. Each surgeon returning data was assigned a postcode for their NHS practice and this was used to create a choropleth map of the workload data based on regions within the UK. Prevalence rates for surgery were then calculated for each region per 100,000 population using mid-2010 population estimates from the Office for National Statistics (ONS). 33 Consultants who did not consider themselves as FAI specialty surgeons were removed from the database. Consultants who considered themselves to be FAI surgeons but were not currently performing this type of surgery within their NHS practice were kept on the database, with reasons for this recorded (e.g. no current funding for the procedure). Anyone who was considered a FAI surgeon but did not want to participate in the study was removed from the database. Data collection was undertaken over a 6-month period between May and October 2012.

In order to triangulate and validate the FAI surgical workload data obtained by survey, NHS Hospital Episode Statistics (HES) data were obtained for procedures undertaken in 2011/12 using Office of Population Censuses and Surveys Classification of Interventions and Procedures-Fourth Edition (OPCS-4) codes within England. OPCS is a procedural classification for the coding of operations, procedures and interventions performed during inpatient stays, day case surgery and some outpatient attendances in the NHS. OPCS-4 is an alphanumeric nomenclature, with a four-character code system. The code system can also be combined to provide further detail. Specific procedure codes for FAI surgery have not yet been established.

In the absence of any established OPCS-4 codes for FAI surgery, the codes currently agreed and applied to FAI surgery within University Hospitals Coventry and Warwickshire (UHCW) were used: Z843, which represents surgery on the hip joint, combined with W844, endoscopic decompression of joint, or W802, open debridement of joint.

Validation of the HES data was undertaken using an independently locally collected database for FAI surgery at UHCW.

International Business Machines (IBM) Statistical Product and Service Solutions (SPSS) Statistics (IBM Corporation, Armonk, NY, USA) version 21 for Windows was used for the statistical analysis. Summary statistics including mean [with confidence intervals (CIs)] and median (with interquartile ranges) values were reported for the data. Differences in workload data between arthroscopic and open surgery were analysed using a Student’s t-test. The level of statistical significance (p-value) was set at 0.05.

Results

There were a total of 193 NHS hospital health boards and trusts in the UK. Of these, 27 did not have an orthopaedic surgical department/service. A total of 2399 cases of surgery were undertaken for FAI over the 12 months. The breakdown of the workload data (both arthroscopic and open) for FAI surgeons who responded to the survey is shown in Table 1.

Of the 100 surgeons returning workload data, 25 did not perform any arthroscopic surgery over the 12-month period and 55 did not perform any open surgery over the 12-month period. The distributions of caseload by surgeon for open and arthroscopic surgery are shown in Figure 2 and the mean and median numbers of cases of FAI surgery per surgeon are shown in Table 2.

FIGURE 2.

Distributions of caseload by surgeon for open and arthroscopic surgery. (a) Individual surgeon workload for FAI treated arthroscopically; and (b) individual surgeon workload for FAI treated with open surgery.

Each surgeon returning data was assigned a postcode for their NHS practice and this was used to create a choropleth map of the workload data based on regions within the UK. Figures 3 and 4 show the number of cases of FAI surgery performed by arthroscopic and open surgery, respectively, in regions of the UK.

FIGURE 3.

Choropleth map of arthroscopic surgery for FAI 2011/12. Number of cases of FAI surgery performed by arthroscopic surgery.

FIGURE 4.

Choropleth map of open surgery for FAI 2011/12. Number of cases of FAI surgery performed by open surgery.

Prevalence rates for FAI surgery using mid-2010 population estimates from the ONS33 are shown Table 3.

Distribution of the numbers of cases per hospital trust is shown in Table 4, combining surgeons where they worked in the same trust.

Trusts with a workload of ≥ 10 hip arthroscopies for FAI per 12 months are summarised in Table 5.

From the initial survey there were 69 hospital trusts in England with at least one FAI surgeon. Of these 69 trusts, 53 had evidence of coded procedural activity from the HES data search. However, the data for both arthroscopic and open surgery were different by five or more procedures in 68 trusts when the two sources of data were compared. Local database results at UHCW showed that 12 arthroscopic and zero open surgeries were undertaken for FAI. The corresponding HES data reported 15 arthroscopic and more than one but fewer than five open surgeries for FAI.

Discussion

A minimum of 120 practising NHS FAI consultant surgeons collectively undertook at least 2399 FAI surgical procedures over a 12-month period. Of these, 1908 procedures (80%) were performed arthroscopically.

Although all health boards and trusts and corresponding clinical directors responded to our enquiries, it is possible that clinical directors are not always aware of all the expertise within their department. New FAI consultant surgeon appointments made after submission of data from the departmental clinical leads would not have been included in this data set, but the workload data for these surgeons are likely to be small given the short tenure and may already be accounted for in work being undertaken by existing surgeons.

The survey data rely heavily on surgeon recall, which is likely to have inaccuracies. However, the alternative of coded procedural data suffers from two major problems:

• Coding of surgical procedures is known to be inaccurate. One study has reported accuracy of 47%. 34 The reason for this is that coding is frequently undertaken retrospectively by staff with no medical training and there are multiple ways of coding the same procedure. 34

• Femoroacetabular impingement surgery has no specific procedural codes and is, therefore, coded using alternative combinations of the OPCS-4 coding system across NHS trusts.

For this reason, the HES data were used to triangulate/confirm locations of FAI surgery rather than provide any robust measure of the quantity of surgery.

Based on the ONS population data, the prevalence of surgery for FAI within the UK is 3.9 per 100,000. This figure represents a conservative estimate for prevalence of surgery for FAI within the UK because

• a total of 20 (17%) surgeons did not provide data

• the survey does not include surgery done outside the NHS.

In the absence of any formal operative coding for FAI, and with FAI surgical registries only recently commencing, this type of prevalence data has not been available to date through other sources. Although all health boards and trusts and corresponding clinical directors responded to our enquiries, it is possible that clinical directors are not always aware of all the expertise within their department. New FAI consultant surgeon appointments made after submission of data from the departmental clinical leads would not have been included in this data set; however, the workload of these surgeons is likely to be low given the short tenure and may all already be accounted for in work being undertaken by existing surgeons.

The results suggest the workload of FAI surgery is not spread evenly by region. There is a suggestion that more surgery is taking place in the South West, London and West Midlands regions per head of population, with marked variations in surgical workload across neighbouring regions, for example West Midlands and East Midlands (5.8 and 0.4 per 100,000, respectively). It is unlikely that these differences are due to regional variations in the prevalence of FAI; they and are more likely to represent regional variation in both surgical expertise and funding for FAI surgery. Surgical treatment of FAI is a relatively new and technically demanding procedure;35,36 therefore, the availability of surgeons with sufficient experience and expertise to undertake such surgery is likely to be limited. This is supported by the high workload among a small number of surgeons nationally. The data have highlighted marked regional variances in surgical workload and it is possible that some patients by the nature of their geography may not have easy access to the care that they would like.

The data presented provide epidemiological data on FAI surgery including those sites with a combined workload of 10 or more arthroscopic FAI procedures over 12 months. These sites may be particularly suited to become recruitment units for a RCT, provided the staff members involved in a RCT are in a position of equipoise.

The survey results provide a conservative estimate for both the number of FAI surgeons currently practising in the NHS and the prevalence of FAI surgery being undertaken. The results also highlight marked regional variances in surgical workload and it is possible that some patients by the nature of their geography may not have easy access to the care that they require. Finally, the results presented can be used to help plan a multicentre RCT based on the FAI surgical workload at each hospital trust.

Introduction

To participate in RCTs, clinicians need to accept at least collective uncertainty or equipoise between treatments, including the possibility that surgery does not work.

Objectives

To explore attitudes to randomisation of patients with FAI among a multidisciplinary sample of clinicians working in the centres performing hip arthroscopy.

Methods

We aimed to recruit a convenience sample of health-care professionals who have a special interest in hip conditions and were likely to see patients with FAI during their clinical practice. Semistructured interviews were carried out with a sample of orthopaedic surgeons, sport doctors and physiotherapists. To stimulate discussion, clinicians were provided with anonymous patient cases that cover the spectrum of patient presentations, including patient history (with duration of symptoms and previous treatments), examination findings and imaging (see Appendix 1). The clinicians were asked to ‘think aloud’ while considering the patient cases and the researcher used prompting questions to facilitate the clinician narrative of these processes. By introducing patient cases, we sought to deconstruct the cognitive processes involved when a clinician considers recruiting a patient to a clinical trial. This was done in order to elucidate the state of individual equipoise that may be influencing a recruitment decision (see Appendix 2).

The interviews were analysed thematically based on the Buckingham–Adams classification model of clinical decision-making. The analysis allowed to identify (1) relevant cues; (2) psychological representation of these clues (i.e. importance in diagnosis); (3) knowledge structures used (e.g. past clinical experience); (4) condition and treatment path inferences; and (5) potential outcomes associated with a particular treatment including introducing the trial. Key messages about what relevant cues to decision-making, information sources, clinical uncertainty and risks, and potential outcomes were considered by the clinicians were identified and discussed within the qualitative team. Recommendations about clinician recruitment and information provision for clinical teams involved in the trial were agreed and presented to the trial management team.

Results

A total of 28 clinicians were interviewed. Eighteen orthopaedic surgeons were identified from our workload survey (see Hip arthroscopy for femoroacetabular impingement in the NHS in the UK: a workload survey) as being high-volume arthroscopic FAI surgeons (> 20 procedures per year). Half of the physiotherapists (n = 6) were identified through a national survey (see Design of a best conservative care treatment protocol for femoroacetabular impingement), which requested the opinions of professionals who have a special interest in hip conditions. The other six physiotherapists and two sports physicians were interviewed when they attended a during an arthroscopic hip surgery conference. A detailed report of the recruitment processes and analysis of these interviews is presented in Appendix 3.

The results showed there was a lack of consensus about how best to treat FAI. The majority of clinicians accepted uncertainty between operative and non-operative treatments. Twenty-six clinicians (90%) believed that they were in equipoise and that a RCT was required to generate superior scientific evidence and guidelines for the care of patients with FAI. These were urgently needed, as they have witnessed increasing numbers of FAI patients in their routine practice. There were various statements supporting the trial; for example, a surgeon said:

The study is a very important one, it has not been done before and we need to do it to justify the role of surgery for this condition.

Surgeon 10

Despite this, five surgeons (36%) and two physiotherapists (10%) showed a lack of active clinical equipoise when faced with real-life case scenarios or discussing involvement with a pilot RCT. One surgeon has a fundamental disbelief in FAI, so that a trial of its treatment lacks relevance for them:

I’m not a massive believer that hip impingement exists, I think it’s over diagnosed.

Surgeon 3

Other clinicians are not in equipoise because they approach surgery for FAI with caution (n = 2); for example, a surgeon, referring to his practice, said:

I’m very careful that I counsel patients about the length of rehabilitation required after surgery, and that it’s definitely only for failure of conservative management.

Surgeon 6

Finally, some surgeons favoured surgery as the optimal treatment for FAI (n = 2), which is the case for the two physiotherapists who were not in equipoise; for example, one surgeon said:

I don’t believe there is any more a question of whether good surgery can relieve [FAI] symptoms.

Surgeon 9

There were seven surgeons who displayed both ‘theoretical’ and ‘active’ equipoise when assessing the case vignettes and these surgeons were invited to act as principal investigators (PIs) in the pilot RCT. Of these seven, there were three who were clearly of the strong belief that a RCT was the optimum solution to improving evidence and the remaining four were permissive.

A major concern for the surgeons (n = 10) was the duration of the trial, needing to balance sufficient length of follow-up to see changes with the potential for deterioration of a patient’s hip during conservative care; for example, one surgeon said:

The main concern with all of this is that if you’re delaying surgical treatment, are they going to progress essentially their arthritic symptoms or cartilage damage by introducing that delay.

Surgeon 13

They felt that the proposed duration of 12 months for the trial was pragmatic and, therefore, acceptable in this respect and would also be long enough to allow the patient to stabilise after either intervention. One surgeon indicated this agreement by saying ‘12 months may be a little long, but I can understand with what we are trying to come up’ (surgeon 11). The surgeons were anxious to know the mechanism by which the outcome of the trial would be assessed and were more comfortable with a patient-reported quality-of-life instrument rather than any other types of measure. All were familiar with and supported the use of either the Non-Arthritic Hip Score (NAHS) or the International Hip Outcome Tool (iHOT-33) as the primary outcome measure for a full trial.

The physiotherapists and sports physicians frequently mentioned two issues: (1) the need for more clarity of the eligibility criteria (i.e. patients who do not need surgery do not take part) and (2) the need for the new conservative care protocol to be more distinctive. For example, a physiotherapist said:

As long as conservative care is standardised and you’ve got a level of expertise by the people giving the conservative care, I think it is very valid.

Physiotherapist 5

The majority of clinicians that assumed patients would prefer surgery and were concerned about patient reactions to an invitation to participate in the study, which meant that patient expectations were not fulfilled. They also thought patient would respond negatively to being randomised. The following quotes illustrated these assumptions:

Patients get to the point where they’ve tried everything and sometimes they are looking for a surgical route, she had failed physiotherapy.

Physiotherapist 4

She’s going to be very unwilling not to have an operation, I would imagine.

Surgeon 3

Finally, clinicians reported potential regional sources of bias in the recruitment. For example, surgeons’ preferences for conservative or operative care could be at play. One surgeon said:

I think if it had been done 3–4 years ago, I think it would have been easier, it was less conviction and most surgeons that were doing it that it was the right thing to do.

Surgeon 14

Although another surgeon explained his own view as:

Of the majority of surgeons offering HA [hip arthroscopy] in the country, I see myself as being a little less invasive and more conservative [. . .] there are surgeons around the country who will operate in patients that I wouldn’t touch with a barge pole.

Surgeon 6

There are also perceived differences in physiotherapy provision. Clinicians reported their concern that local physiotherapy may not be of equal quality across the centres:

Are you suggesting a remote access physio who gives them a programme and liaises with the local physios? Because we don’t have specialist enthusiastic physios in our area which would be quite as knowledgeable as the ones you have in Coventry.

Surgeon 7

Discussion

Significant interest was evident in the clinical community for a RCT comparing hip arthroscopy and conservative care. Even though not all clinicians were in equipoise, almost all recognised uncertainty and felt that a RCT would help to guide their practice in future. Most surgeons were prepared to randomise patients.

The qualitative methods we described provided a novel approach to determining a more complete assessment of clinical uncertainty and equipoise. The results suggest that, while many clinicians might think they were in equipoise (theoretical equipoise) their actions and management decisions did not always support this position (active equipoise). It was likely that where there was a large discrepancy, recruitment of patients to a RCT was going to be challenging. Therefore, decisions about participating sites and PIs were informed by the results of this study. It also helped to provide guidance of study design features including follow-up duration, outcome measures and the need for distinctiveness of the best conventional care arm.

Introduction

To date, no RCTs have been undertaken for FAI surgery and no previous eligibility criteria exist to guide a pilot. In addition, as a relatively new procedure, there are a variety of surgical techniques for arthroscopic FAI surgery and, therefore, it is necessary to develop and define a protocol of operative care in order to ensure that patients receive the same standard intervention and that the fidelity of this intervention can then be measured.

Objectives

• Establish appropriate eligibility criteria for a RCT.

• Establish a protocol of operative treatment for arthroscopic FAI surgery.

Methods

Researchers PW and DG developed draft eligibility criteria and an operative protocol based on their own experience of arthroscopic FAI surgery and available published literature. Sixteen hip arthroscopy surgeons, recognised as international experts in the field, were then individually invited to comment on these provisional documents. These surgeons were from the Multicenter Arthroscopy of the Hip Outcomes Research Network (MAHORN) (n = 12) and from a group who attended the American Academy of Orthopaedic Surgeons Research Symposium on Femoroacetabular Impingement in Chicago, IL, USA (n = 4). All surgeons approached agreed to participate. The feedback they provided was then used to modify the two documents. The eligibility criteria and protocol were recirculated to the 16 experts for further comment. The criteria and protocol were then discussed with a further sample of 14 UK specialist hip surgeons with experience of treating patients with FAI and likely collaborators for a full RCT. A final version of both documents was agreed and then used in the pilot RCT. The two documents continued to be evaluated in the light of feedback from the recruiting sites during the pilot RCT.

Results

The draft eligibility proposed that patients be included if:

• they are aged 18–50 years

• they have symptoms of hip pain: they may also have symptoms of clicking, catching or giving way

• they show radiographic evidence of pincer- or cam-type FAI on plain radiographs and cross-sectional imaging37

• the treating surgeon believes that they would benefit from arthroscopic FAI surgery

• they are able to give written informed consent

• they are able to participate fully in the interventions.

Patients would be excluded from participation in this study if:

• they have previous significant hip pathology such as Perthes’ disease, slipped upper femoral epiphysis or avascular necrosis

• they have had a previous hip injury such as acetabular fracture, hip dislocation or femoral neck fracture

• they already have osteoarthritis, defined as Tönnis grade of > 1,38 or more than 2-mm loss of superior joint space width on anteroposterior pelvic radiograph8

• there is evidence that the patient would be unable to participate fully in the interventions, adhere to trial procedures or complete questionnaires, such as cognitive impairment or intravenous drug abuse.

Initially, 11 out of the 16 surgeons agreed with these eligibility criteria. Others suggested the following modifications, which were agreed by the whole group:

• Age range changed to include all patients ≥ 16 years. Some surgeons felt that patients between 16 and 18 years of age were an important part of their practice and that for these patients treatment effects would be similar to older adults. There was not thought to be any rationale for a maximum age limit; older patients may be, but need not necessarily be, excluded by the criterion ‘the treating surgeon believes that they would benefit from arthroscopic FAI surgery’.

• Radiographic evidence of FAI was further defined to include an alpha angle of > 55 degrees or a lateral centre–edge (Wiberg) angle of > 40 degrees.

The draft operative protocol was:

• general anaesthetic with muscle relaxation

• supine or lateral patient positioning

• operating table with facility for traction and allowing range of movement testing

• arthroscopy of central compartment

• arthroscopy of peripheral compartment working with one of the following: intact capsule, capsulotomy or capsulectomy

• ability to undertake bony surgery to correct abnormalities on both the femoral head neck junction and acetabular side of the hip joint

• ability to undertake soft tissue repair and/or debridement to the labrum and/or articular cartilage

• ability to record with photos the intraoperative findings and solutions.

A total of 13 out of the 16 surgeons agreed with the draft operative protocol. Three surgeons proposed changes, which were agreed by the whole group:

• the entire acetabular labrum should be examined

• the entire articular surface should be examined

• confirm that FAI has been relieved using either range of movement testing or an image intensifier

• need to document intraoperative complication and their solutions.

Discussion

There was ready agreement on eligibility criteria for a trial, suggesting that surgeons have a clear idea who may benefit from hip arthroscopy for FAI. There was recognition that there are many subtleties of FAI morphology, but surgeons were happy to be pragmatic and ‘lump’ all of these into a single diagnosis. There may be differences in treatment effect according to type of FAI (some surgeons mentioned their sense that patients with cam-type FAI do better after surgery than those with pincer-type) and this raised the possibility of stratification or an a priori subgroup analysis in a full trial. These criteria define a generalisable sample representative of the patients currently receiving hip arthroscopy for FAI.

The operative protocol was broad enough to accommodate the variations of arthroscopic FAI surgical technique being used throughout the UK. However, the protocol retains the key steps regarded by the expert surgeons as essential to a successful hip arthroscopy for FAI.

We considered whether or not to protocolise the postoperative rehabilitation programme for the group randomised to surgery prior to the pilot trial. This is a pragmatic trial and we believe that the best option is for postoperative rehabilitation to reflect usual care as closely as possible. Postoperative rehabilitation varies, often from surgeon to surgeon and even within the same orthopaedic service. If we were to develop a postoperative rehabilitation programme that is similar to the personalised hip therapy (PHT) programme, this would change the question for the trial (and, therefore, it would no longer address the original commissioned call from the HTA programme). The question would then become ‘Does surgery provide additional benefit to a package of physiotherapy-led care?’ Our preference was to allow postoperative care to be offered as per usual practice for those randomised to surgery in the pilot trial and to measure that to allow for assessment of potential confounding.

Introduction

Although conservative care is being used to treat patients with FAI,11 we had previously performed a systematic review that showed that detailed guidance and evidence on how this care should be delivered are not available. 13 This review suggested that non-operative care for FAI would typically be led by physiotherapists. Of 53 published articles, only four were empirical investigations of non-operative care for FAI and only one study39 provided both an experimental evaluation of treatment and an explicit description of the treatment protocol delivered by physiotherapists.

When there is a lack of evidence to guide care, it is appropriate to use a consensus-gathering approach in order to develop and rationalise best practice. 40 Murphy et al. 40 summarised three formal consensus-gathering techniques which have been used to guide health care in other subject areas:

• Delphi method – involves participants receiving two, three or more sequential rounds of questionnaires and responding to ‘cues’, that is, statements that provoke decision-making based on responses from previous rounds. Results are aggregated and reviewed for agreement. 41

• Nominal group technique (NGT) – involves a process of generating ideas which are then either accepted or rejected by group members. 42,43

• Consensus Development Conference (CDC) – requires a group of individuals to attend a conference in which evidence is presented to them by experts. 44

Objectives

• To develop a consensus on a best conservative care treatment protocol for patients with FAI that was both deliverable within the NHS and could be used within a RCT of best conservative care compared with hip arthroscopy surgery for FAI.

• To agree, through patient involvement, the most appropriate name for the best conservative care treatment protocol.

Methods

The protocol proposed by Emara et al. 39 (Table 6) was used as the starting point for a best conservative care treatment protocol for FAI. This was developed using Delphi and NGT consensus-building methods, guided by available evidence and guidance from the MRC for developing a complex intervention. 45

A core study group was formed to oversee the development, evaluate information gathered and provide a layer of NGT consensus to support the Delphi process. The core study group comprised two senior musculoskeletal physiotherapists with an interest in managing patients with FAI (David Robinson and Ivor Hughes), a senior academic research physiotherapist (NF) and an orthopaedic surgeon (PW).

When we began this study, we did not know which physiotherapists were directly involved in the management of patients with FAI, nor did we think it likely to be efficient to simply sample the physiotherapy profession in the UK at random. We therefore took a targeted approach to sampling, using networks of physiotherapists most likely to be involved in the management of this patient group. National advertisements were placed in the orthopaedic, rheumatology, pain and manual therapy electronic networks through the interactive, electronic Chartered Society of Physiotherapy (CSP) communication system in the UK (iCSP) and in the CSP’s Frontline magazine (twice-monthly magazine posted to approximately 52,000 CSP members in the UK). The adverts invited UK physiotherapists to help develop a consensus for a best conservative care treatment protocol for FAI. Electronic invitations were also sent to physiotherapists in the USA and Australia known to members of the core study group through previous collaborative work on FAI. To encourage a process of ‘snowball sampling’ within the international community, these therapists were encouraged to invite colleagues with experience and interest in managing FAI to join in the consensus development process.

Each physiotherapist was given the first protocol39 with a questionnaire. The physiotherapists were asked whether they agreed or disagreed with the proposed programme for the conservative treatment of FAI patients and, when appropriate, to provide comments and suggestions for improvement.

Results from the each round of Delphi consensus were tabulated by the core study group, and additional comments and treatment strategies suggested by the respondents were grouped into themes. An agreement level of ≥ 50% for this Delphi consensus technique was used. If no consensus was evident from the survey the core study group refined the protocol in light of the available feedback using a NGT-type approach. The refined protocol was then recirculated to the physiotherapists taking part in the Delphi consensus process and the cycle repeated until a consensus of ≥ 50% was achieved.

This protocol was then discussed with our panel of expert patients (described in section Patients’ attitudes towards randomisation and design of patient information material for a randomised controlled trial) in individual interviews. They expressed the view that the name ‘best conservative care’ did not fully express (to patients) the intent and content of the protocol. Previous qualitative research has highlighted the importance of naming treatments in order to improve uptake and compliance, in particular when being used in RCTs. 46 The expert panel was asked to suggest a suitable name for this best conservative care treatment protocol.

All physiotherapists identified as likely to be best conservative care providers in the pilot RCT were asked to detail exercises that would allow them to deliver the protocol. The exercises were then ranked and the most popular were included as a database resource (exercise template) to be used alongside the best conservative care protocol. In the early phases of recruitment to the RCT, a workshop (CDC methodology) was held among the physiotherapists delivering care, to refine the protocol. The physiotherapists attending the workshop were asked to share their experiences of delivering the protocol and make any suggestions for further amendments.

In the pilot RCT, all physiotherapists delivering the protocol were asked to complete case report forms for each patient. This included details about the number, nature and duration of the patient contact, and details of the exercises prescribed to each patient were recorded.

Results

Developing an exercise template and testing the protocol

Twelve physiotherapists were initially identified as likely to be providers of the best conservative care protocol during the pilot RCT in their hospitals. They suggested suitable exercises for core component 4 and the most popular 21 exercises are shown in Appendix 4.

A workshop (CDC-type methodology) was held after 42 patients were recruited to the RCT and 21 patients were allocated to and receiving the PHT protocol. Eight physiotherapists (out of 12 physiotherapists participating in the RCT) from seven recruiting centres attended the workshop to review the content and delivery of the protocol. Collectively, the physiotherapists were treating 18 patients within the RCT. The physiotherapists agreed that the protocol worked well but felt that they wanted to change the number of treatment sessions and the overall duration of the protocol, in order to ensure that they were able to deliver best care. As a result, one change was made to the protocol, which allowed a minimum of six and a maximum of 10 contacts over a 6-month period. In addition, the physiotherapists recommended that three further exercises should be added to the selection of 21 within the exercise template (see Appendix 4). The physiotherapists agreed that no further amendments would be needed to the protocol.

Discussion

Despite the increase in attention and recognition of FAI as a source of hip pain in young adults, there has been very little published information about appropriate conservative treatment approaches. Our aim was to develop an agreed high-quality best conservative care and physiotherapy-led treatment protocol for patients with FAI that could be used within a RCT. We combined results from a systematic review, Delphi consensus surveys with FAI physiotherapists, NGT consensus methodology, relevant literature on effective conservative care for other musculoskeletal pain conditions and the experiences of physiotherapists treating FAI patients within a RCT (CDC methodology), in order to develop the agreed treatment protocol, referred to as PHT.

In summary, PHT comprises four ‘core’ elements designed to be offered over a maximum of 6 months: (1) patient assessment; (2) patient education and advice; (3) help with pain relief; and (4) engagement with a supervised individualised exercise-based hip programme. A minimum of six and a maximum of 10 treatment contacts should be provided by the supervising physiotherapist over the 6-month period.

The protocol followed MRC guidance for the development of complex interventions and is based on theory when possible. 45 Research has already shown that exercise is an effective treatment for many types of musculoskeletal pain51,52 and has identified that exercise-based programmes can produce similar improvements in symptoms to surgery. 53 The PHT protocol provides guidance to other clinicians and researchers in an area where evidence and guidance are very limited.

Introduction

Producing high-quality patient information sheets (PISs) and consent forms that are clear, address the relevant patient concerns and are presented in an easy to assimilate format is essential for successful recruitment to trials. In order to increase the numbers of patients randomised for the proposed trial, a qualitative study was set up to investigate concerns that patients may have and how best to inform patients of this trial.

Objectives

• Explore patients’ views about operative and conservative treatments for FAI and the acceptability of randomisation between them.

• Develop PISs for the RCT.

Methods

A panel of expert patients was established from among patients who had been treated for FAI at the UHCW NHS Trust. The panel was chosen by purposive sampling in order to include a range of age, sex, disease severity, activity and socioeconomic status as well as those who had received operative and conservative care. An invitation to be part of an expert patient panel was sent to patients and anyone who replied to the invitation and signed the consent form was asked if they were willing to participate in one-to-one interviews. Once the expert panel member declared their willingness to be interviewed, the researcher (AR) contacted them and explained the qualitative study. The PIS and consent form was sent by post and, when the consent form was returned signed by the participant, an appointment for the interview was set up, whether face to face at the hospital or by telephone.

A draft version of a PIS for a RCT comparing hip arthroscopy with best conservative care for FAI was prepared. Panel members were given this draft version and a questionnaire to complete about its readability. Panel members then had a semistructured interview (with AR), either face to face or over the telephone, about their experiences with FAI, the treatment they had received and the answers they gave to the PIS questionnaire (see Appendix 5).

Initially, we had planned to give questionnaires to 50 patients and to interview 15 of these. However, it emerged that the one-to-one interviews provided much richer in-depth information than administering surveys alone and that this allowed data saturation to be achieved.

Transcripts of interviews were prepared and thematic analysis was used to extract key messages about patient experiences of hip impingement and treatments, acceptability of the treatment options, the understanding of randomisation, and advice on how to improve the trial participant information sheet. AR coded the transcripts using NVivo (QSR International, Warrington, UK). Agreeing and contrasting views were identified for each area of interest (e.g. randomisation). After discussing the structure and content of the results, the qualitative recruitment intervention (QRI) team agreed on some recommendations which were presented to the trial management team. Further discussions took place in trial management meetings and, once agreement was reached regarding the recommendations, changes to the trial PIS and recruitment strategies to be tested in the pilot were implemented.

Results

A total of 18 FAI patients were part of the expert patient panel; 14 of them received operative treatment and four had managed the condition with physiotherapy and other non-operative interventions (e.g. steroid injections). A detailed report of the results of these interviews is presented in Appendix 2.

Patient information sheet assessment

The expert patient members stated that the PIS presented a balanced view of the two treatments that were being compared. The language and illustrations used in the document were appropriate and accessible. Furthermore, patients found and understood easily the information about why it is important to treat hip impingement, the purpose of the study and what operative care involves and its risks. One patient mentioned there was not much information on the risks of not treating hip impingement. Patients did not find satisfactory the information related to conservative care risks, what happens after the study finishes and 50 : 50 treatment allocations. These items were related to the explanations about the trial procedure.

Patients suggested that the PIS would benefit from adding a rationale for offering physiotherapy for a morphological condition such as FAI. They also commented that patients needed to be reassured they would continue with the same consultant and health-care team, and that they needed to be given more details about what happened after the end of the trial. Additionally, some members of the panel said patients tended to look up FAI and available treatments on the internet, so links to trusted websites were suggested. Information about follow-up, details about possible benefits of the best conservative treatment and trusted website links were added to the PIS. The words ‘trial’ and ’50 : 50’ were removed and replaced by ‘study’ and ‘allocated to one treatment’, respectively.

Finally, the expert patient panel emphasised the importance of making patients feel they are receiving personalised care for their hip. They said that being given enough time to ask questions should be a priority during the recruitment consultations and patients should be reassured they would receive adequate treatment independently of how it is chosen. They thought the name ‘personalised hip therapy’ corresponded to these patient needs.

A new version of the PIS was developed that addressed these comments. It was reviewed and approved by the patient and public involvement (PPI) member of our research team and is shown in Appendix 6.

Discussion

Patients were prepared to take part in research to identify the best treatment for FAI and were even enthusiastic about a trial. They expressed concerns about the language of random allocation and wanted to be assured that treatment would be personalised for them. Patients felt that these concerns could be addressed by modifications in the PIS, which we subsequently tested in the pilot RCT.

We recognise that this study was of expert patients who had already been treated for FAI. Their assertions that they would have been prepared to be randomised, when considering the question retrospectively, may not be representative of the behaviour of patients presented with a new diagnosis and making a decision that would actually affect their treatment. A prospective study of patients in this situation is presented in the pilot RCT. However, this was sufficiently reassuring to support performing that RCT, and comments from these patients substantially influenced the design of that pilot.

Design

A pragmatic multicentre, parallel group, pilot RCT of hip arthroscopy compared with best conservative care for FAI with a QRI was undertaken at 10 NHS centres within the UK.

Participants

Patients were eligible to participate in the study if:

• they were aged ≥ 16 years

• they had symptoms of hip pain – they may also have had symptoms of clicking, catching or giving way

• they showed radiographic evidence of pincer- or cam-type FAI on plain radiographs and cross-sectional imaging37

• the treating surgeon believed that they would benefit from arthroscopic FAI surgery

• they were able to give written informed consent and to participate fully in the interventions.

Patients were excluded if:

• they had previous significant hip pathology such as Perthes’ disease, slipped upper femoral epiphysis or avascular necrosis

• they had a previous hip injury such as acetabular fracture, hip dislocation or femoral neck fracture

• they had osteoarthritis, defined as Tönnis grade of > 1,38 or more than 2-mm loss of superior joint space width on anteroposterior pelvic radiograph8

• there was evidence that the patient would be unable to participate fully in the interventions, adhere to trial procedures or complete questionnaires, such as cognitive impairment or intravenous drug abuse.

Participants were recruited from among patients presenting to young adult hip clinics in each of the following 10 NHS centres:

• UHCW, Coventry, UK

• Yeovil District Hospital, Yeovil, UK

• Royal Devon and Exeter Hospital, Exeter, UK

• Royal Orthopaedic Hospital, Birmingham, UK

• Frimley Park Hospital, Frimley, UK

• Royal Cornwall Hospitals, Truro, UK

• Epsom and St Helier University Hospitals, Epsom, UK

• Guy’s and St Thomas’ Hospitals, London, UK

• St Bartholomew’s and Royal London Hospitals, London, UK

• University College Hospitals, London, UK.

This was an increase in the number of sites originally planned and included some changes of sites. During the pre-pilot phase the surgeons from all of the eight potential sites listed in the protocol v 12.0 12/02/2012 were interviewed to determine their equipoise and willingness to take part in the pilot trial. The interviews suggested that some surgeons would not be suitable recruiters/PIs. However, interviews with surgeons from other potential sites were more promising and as a result they replaced some of the original sites.

One potential site declined participation as it wanted to take part in another study involving patients diagnosed with FAI and felt it could not offer both studies to patients.

Recruitment

Possible patients with FAI (adults with hip pain and without a prior diagnosis of osteoarthritis) were identified by collaborating surgeons from referral letters. These patients were then invited to a diagnostic consultation with one of the collaborating surgeons. Prior to their appointment, patients were approached to seek consent for audio recording of their clinic consultations.

Surgeons assessed patients as usual, taking a history, examining the patient and performing appropriate imaging investigations. Patients in whom a diagnosis of FAI was made, and who met the eligibility criteria, received a ‘diagnostic consultation’ including a description of the condition from their surgeon and an explanation that there are two possible treatments. This diagnostic consultation was recorded where consent had been given. Patients were provided with a PIS about FAI and the pilot RCT.

Patients were then invited to attend a ‘recruitment consultation’ by a trained research associate. Typically this occurred on the same day and during the same clinic. During the recruitment consultation, information was provided about FAI and its possible treatments and about the pilot RCT. Patients were given an opportunity to ask questions. This recruitment consultation was recorded where consent had been given. Patients were then invited to give their consent to become participants in the RCT. Patients who wished to take more time to consider were given an opportunity to do so. Those who agreed to take part completed baseline questionnaires and were then randomised to one of the two treatments.

Treatment allocation

Participants were randomly allocated to arthroscopic surgery or best conservative care using 1 : 1 secure centralised telephone randomisation provided by Warwick Clinical Trials Unit. Patients were usually informed of their allocation at the recruitment consultation and plans for delivery of the intervention made at this point. Participants were allowed to withdraw from the trial at any time without prejudice. Participants who decided to have the treatment to which they were not randomised were followed up wherever possible and data collected as per the protocol.

Interventions

Participants were treated according to protocols summarised below.

Outcome assessment

Although the primary objective was to determine recruitment to a pilot RCT, patients were followed up as though they were part of a full RCT and the study conducted as though it was an internal pilot RCT. The following outcome measures were used:

• hip pain and function – NAHS and the iHOT-33 (see Chapter 1)

• activity level – University of California, Los Angeles (UCLA), activity score (a reliable and validated patient-reported activity grading system)66

• health-related quality of life – Short Form questionnaire-12 items (SF-12)67 and EQ-5D68

• a resource-use questionnaire.

These measures were taken at 3, 6 and 12 months after randomisation using a postal questionnaire (see Appendix 7). Questionnaires were posted and recipients were then followed up twice by telephone.

Concealment of treatment allocation

Patients, surgeons and physiotherapists were not blinded to the treatment allocation. The patient-reported functional outcome data were collected by post and entered onto the trial central database by a research assistant who was blinded to the treatment allocation. The statistical analysis was also performed blind.

Consent

Written informed consent for both consultation recording and involvement in the pilot RCT was undertaken by a clinical researcher or treating surgeon delegated and trained by the central research team.

Sample size

The pilot RCT was not powered to estimate a treatment effect, but rather to achieve a reasonable CI around the estimate of the recruitment rate, measured as the proportion of eligible patients who consent to participate and be randomised. Previous studies in orthopaedic surgery of operative compared with non-operative treatments have achieved recruitment rates of around 30%, so we modelled around this. Table 10 shows 95% CIs around various recruitment rates for a range of scenarios. Approaching 60 patients to invite them to take part in the trial would allow a 95% CI of 18% to 41% if the true recruitment rate was 30%. We judged this to be a good balance between cost and duration of the pilot RCT and the required precision of the estimate.

Therefore, we approached 60 eligible patients across 10 centres in order to estimate recruitment with sufficient precision to plan a full RCT.

Adverse events and serious adverse events

Adverse events (AEs) and serious adverse events (SAEs) not already identified locally and reported from the recruiting site were determined by a patient questionnaire at 6 weeks and 3, 6 and 12 months. Each recruitment site that discovered a SAE was obliged to report details by fax to the Clinical Trials Unit within 24 hours of the investigator becoming aware of them. Once received, causality and expectedness were confirmed by the chief investigator. All participants experiencing SAEs were followed up as per protocol until the end of the trial.

Fidelity of treatments

The quality of the evidence produced by a full RCT depends on the fidelity with which the two treatments being compared are implemented. Therefore, the following measures to determine treatment fidelity were tested in the pilot RCT.

Economic analysis

One purpose of the feasibility study was to test measurement approaches and data collection systems that would facilitate an economic evaluation to be conducted as part of a full RCT. A patient-completed economic questionnaire was designed to capture health and social service resource use and any costs borne by the patient, including time and travel costs incurred in the receipt of care, out-of-pocket expenditures and time off work and usual activities. The economic questionnaire also included the EQ-5D and SF-12 health-related quality-of-life measures that can generate health utilities for economic evaluation purposes. The economic questionnaire was completed by study participants at baseline and, subsequently, at 3, 6 and 12 months. Information extracted from the resource-use component of the questionnaire was validated by cross-checking data provided by the first 10 patients concerning contacts with hospital services (inpatient and day case admissions and outpatient visits) at baseline in the lead site against electronic patient records available in the lead site. The validation process did not extend beyond local hospital services and, therefore, did not include contacts with primary care, other community services or contacts with other hospital services.

Statistical methods

Baseline characteristics (e.g. age and sex) are presented for both treatment groups (hip arthroscopy and PHT) to assess balance after randomisation; continuous data (e.g. age) were summarised by means and SDs and categorical (binary) data by counts and percentages. The distributional properties of the baseline scores were assessed visually using box plots. For the 42 patients recruited into the study, all data were complete (i.e. there were no missing data). Only 31 participants had completed 12 months of follow-up at the time of writing (February 2014), so only baseline scores are presented; outcome data will be described in future publications.

Summary of changes to the project protocol

Research Ethics Committee (REC) approval was received on 15 February 2012 for version 12.2 of the protocol. Following the original submission, a further three amendments have been made to the protocol and REC approval has been received for all three (listed below).

Trial approval and procedures

The trial was approved by National Research Ethics Committee (ref: 11/WM/0389) and by national and local R&D committees without changes and registered with the current controlled trials database under reference number ISRCTN 09754699. We identified several problems in obtaining local R&D approval and in setting up the trial across the 10 sites. These problems were different at each location, but were almost all solved. Further details are given in Appendix 9. Local R&D approval at one of the 10 sites was so delayed by local procedures that the centre opened only after target recruitment for the pilot RCT was completed. Once approval had been granted, we did not identify any major problems in the application of trial procedures at any of the sites.

Recruitment

A total of 60 eligible patients were approached for recruitment to the pilot RCT across nine centres during almost 10 months. One centre (site 10) was delayed in being set up and did not attempt recruitment before 60 approaches had already been made. Of these 60 patients, 42 consented to take part in the pilot RCT; therefore, recruitment was 70% (95% CI 58% to 81%). During this period a total of 151 patients were screened for the study. Overall recruitment activity is shown in Figure 5 and recruitment at individual centres is shown in Table 11. Note that centres opened for recruitment at different times and recruited for different periods. Reasons for these differences in uptake were explored in the QRI and they mainly related to logistical issues such as difficulties in setting up funding sources and assigning recruiters to attend clinics attended by potential participants (see Chapter 4).

FIGURE 5.

Recruitment during the pilot RCT.

Participant demographics were as expected, participants typically being young (mean age 35 years) and active (50% had a UCLA score of 1, 2 or 3, indicating regular participation in sport). Table 12 and Figure 6 show patient characteristics and baseline outcome scores by treatment group; there were no significant differences between those allocated to hip arthroscopy and those to PHT.

FIGURE 6.

Box plots of functional hip scores. (a) iHOT-33; and (b) NAHS, with means and 95% CIs.

Follow-up

This pilot study focused on estimating the recruitment rate and was not designed for estimation of a treatment effect. The majority of the participants had not reached final follow-up by the time this report was prepared. However, only 2 out of 42 participants did not receive their allocated treatment (one decided not to proceed with surgery and one asked to have surgery after beginning PHT), and follow-up rates were 98%, 94% and 100% at 3, 6 and 12 months, respectively. A Consolidated Standards of Reporting Trials (CONSORT) diagram for the pilot RCT is shown in Figure 7.

FIGURE 7.

The CONSORT diagram for pilot RCT. a, Follow-up rates reflect progress of patients as of 1 February 2014.

Procedures to ensure fidelity of interventions

Economic analysis

We found a minimum of 98% completeness for all aspects of the economic questionnaires (encompassing resource use and health-related quality-of-life measures) at all follow-up points (Table 13). During the validation process, no discrepancies were found between use of hospital services reported by the patient and the corresponding electronic records.

Qualitative research found the economic questionnaire to be acceptable among patients at the lead site and that it could be completed by patients without any extra support from a research assistant.

We were able to establish estimates of unit costs for the majority of relevant resource inputs from a range of secondary sources, including national cost compendia such as the Patient Social Services Research Unit Costs of Health and Social Care and the British National Formulary. If unit costs were missing, these could be estimated using primary research methods in consultation with the lead site’s finance department.

Methods

A comprehensive process of logging of potential RCT participants through screening and eligibility phases was put in place in order to ensure compliance with the CONSORT checklist. These data were made available to the qualitative researcher (AR) on a regular basis to facilitate monitoring recruitment. These data and interviews with the different recruitment teams helped to develop a flow chart of the most likely patient pathways.

Results

Figure 8 presents a flow chart summarising the patient pathway. It was important to assess patient pathways in relation to their complexity and compliance with the protocol as well as variation between centres. This information provided indications of particular points where patients could potentially be ‘lost’ from the RCT.

FIGURE 8.

The FASHIoN pathway through eligibility and recruitment.

Methods

In-depth, semistructured face-to-face interviews were conducted with members of the TMG, including the chief investigator and those most closely involved in the design, management, leadership and co-ordination of the trial.

Informants were asked about the background, development and purpose of the RCT; their knowledge of the evidence and their own opinion or equipoise; their role in the trial; and their expectation of the pathway through eligibility and recruitment. They were also asked to provide a short verbal summary of the RCT for the interviewer, as if AR was a patient.

Interview topic guides were used to ensure similar areas were covered in each interview, based on those used in previous studies,27 but also encouraging the informants to express their own views about the RCT and any recruitment challenges expected or experienced (see Appendix 10).

Transcripts of the interviews were analysed thematically by AR, using techniques of constant comparison and case study approaches. This involved detailed coding and then comparing emerging themes looking for shared or disparate views among TMG members. The coding was carried out using the qualitative data analysis software NVivo. The initial coding (AR) was checked by two other QRI researchers (AA and JD) and inconsistencies resolved by discussion. Detailed descriptive accounts of the themes and cases were produced.

Results

Ten interviews were conducted with members of the TMG. Analysis of these and observation of the TMG meetings showed the group was formed of four subteams with specific functions:

1. orthopaedic surgeons and senior physiotherapists in charge of designing the interventions, recruiting participants and offering a critical view of the research procedures

2. the trial administration and finance control group dedicated to data management, liaison with other agencies, and finance planning and expenditure

3. expert statisticians in charge of designing and planning statistical analysis

4. qualitative researchers in charge of the QRI.

Members of each subgroup had at least three different levels of involvement with the research. The first one was with the day-to-day running of the trial; at a second level, senior members supervised tasks and made major decisions (e.g. TMG meetings); and, finally, some members of the team remained distant from the trial procedures and provided a critical assessment of the trial and/or consultancy in particular aspects of the research (e.g. statistical analysis).

The team members showed an adequate knowledge of the trial protocol according to their level of involvement. Clinicians were the most knowledgeable as well as those involved with everyday activities, which was as expected. Task delegation within this multidisciplinary team prevented recurrent role conflicts and helped the team to develop ‘collective equipoise’. This term refers to the tendency of the team members to challenge each other’s expectations about the results of the RCT without favouring either of the two treatments tested. For example, although it was likely the orthopaedic surgeons running this trial tended to advocate for the operative treatment of FAI, they had a strong background in research and, therefore, were able to suspend clinical judgement and achieved a balanced view of the treatments that were compared, as illustrated by the following quote:

I distinguish there between my sort of gut feeling and my instinct. My instinct is that it works, surgery works [. . .] However, I do recognise that I have also seen it not work and I recognise that the evidence that supports it working, even the evidence I’ve generated myself, is methodologically weak and therefore I think that a scientific and objective perspective of the evidence [. . .] I am therefore an equipoise so I’m able to hold two views on it. And obviously, I mean the equipoise view is the one that I use when I’m involved in a trial.

TMG01

As a result, the team sought collaboration with physiotherapists at national and international levels and developed a credible conservative care protocol. Additionally, the TMG included professionals without expertise on hip arthroscopy or physiotherapy that facilitated equipoise.

A major concern for the TMG was the multicentre aspect of the pilot. They mentioned challenges such as difficulties implementing and overseeing procedures in other centres, delays in setting up, low numbers of eligible patients and lack of equipoise in research teams. The following quotes exemplified these concerns:

I know that the start of some of the sites has been a bit delayed but that is not unusual nor is it surprising given the complexities of setting up research in different sites.

TMG05

I’ve got major doubts [the study will be successful]. I think it was entirely the right thing to do it as a feasibility study, I have serious concerns that we won’t be able to recruit outside of this centre [. . .] Finding that surgeons and patients at the same time have both got equipoise is just going to be very difficult.

TMG02

Among the suggestions collected from the group to address the concerns were:

• regular visits to the centres by the PI and other TGM members to keep momentum

• delivery of a slick and easy-to-implement recruitment process in order to be the least disruptive to routine clinical practice

• providing frequent and comprehensive training to recruiters helping them to develop skills and confidence (e.g. role playing, FAQ, peer support, etc.)

• modifying the support to teams in other centres according to their research experience

• setting recruitment targets and engendering a healthy competition between centres

• sending regular newsletters providing information about recruitment.

Methods

In-depth, semistructured face-to-face interviews were conducted with clinicians and research associates tasked with recruiting patients in visits to sites soon after they had started to recruit patients for the pilot RCT.

Informants were asked questions about their knowledge of the evidence for the treatment of FAI and their personal views about equipoise; the recruitment pathway; how they feel the protocol fits their clinical setting; and any adjustments they thought were needed to the trial procedures. They were also asked about the audio recording of their consultations with patients, with a view to discussing any discomfort or perceived difficulty with this.

Again, interview topic guides were used (see Appendix 10) and informants were encouraged to express their own views about the RCT and any recruitment challenges expected or experienced.

Analysis was as for the TMG interviews, but themes were also explored to examine differences or similarities between the TMG members and specialist clinicians and recruiters, and within or between centres or clinical specialties.

Results

Twenty-one interviews of clinicians and research associates were performed at eight participating sites. Common concerns and good practice examples were identified across the transcripts and were organised in the following themes.

Methods

The chief investigator and TMG decided that the meeting in which randomisation was discussed, the ‘recruitment consultation’, as well as the prior appointment when the patient was told their diagnosis of FAI, ‘diagnostic consultation’, represented pivotal points of the recruitment to this trial and, therefore, sites were asked to record these two consultations for each potential trial participant.

The importance of audio recording discussions about RCT recruitment was emphasised at site initiation visit. All participating centres in the pilot RCT were asked to audio record recruitment and diagnostic consultations when appropriate and feasible. One main point of contact (usually the research associate) was identified per centre and digital audio recorders were provided. PISs and consent forms for audio recording and instructions for the operation of the recorder, dictation of patient/recruiter/recording identifiers, naming and transferring of the recording to the computer and then to the QRI team were provided to centres. Clinicians were assured the feedback to them was going to be confidential and positive (not critical). Prior to their consultation with the surgeon, patients were approached and the PIS for the audio recording of consultations was given to them to read. Once their questions were answered and if they agreed to the audio recording, patients were then asked to sign a consent form.

Audio recordings of consultations were analysed following thematic analysis and some of the techniques of focused conversation analysis pioneered in previous studies (Donovan et al. 2003;26 Donovan et al. 200927). The analytical techniques were used to identify and document aspects of informed consent and information provision that were unclear or disrupted or which hindered recruitment. The content of the appointments was evaluated, including what basic content was covered, the order of presentation of RCT arms and other treatment options, time spent on interventions and time spent describing both the RCT design and the randomisation process. Furthermore, an assessment was made as to whether or not the recruiter listened to patients’ concerns and addressed them and also the degree to which there was evidence that the participant understood the key issues of equipoise, randomisation, participation in the RCT, the option to choose their treatment and the option to withdraw from the research at any time.

AR documented these details and provided an account for JD and AA. When at least three recordings per recruiter had been analysed, the QRI researchers decided what confidential feedback would be given in order to promote research associates’ communication skills development and recruitment effectiveness. Issues to be fed back to the RCT chief investigator/TMG, or to be used anonymously in training programmes, were discussed and defined.

Results

Eighty-seven diagnostic and recruitment consultations relating to 60 individual patients were recorded and analysed. A subset of eight successful and 13 unsuccessful recruitment consultations were studied in detail in order to understand communication patterns that were linked to improved recruitment rates and which could be repeated in future patient approaches. The remaining audio recordings were used to validate the best recruitment practice model (see Figure 9) and create personalised feedback for recruiters who had submitted the recordings.

Qualitative recruitment intervention support focused on three sites where recruitment was comparatively low (42–66%).

Common problems identified were:

Unbalanced presentations of treatment options, for which surgery has been presented at greater length and more favourably than either choosing conservative care or participating in the RCT (surgeons tend to talk most about what they are most familiar with). The following quote illustrated this information unbalance:

There are other centres and other people have shown physiotherapy to be very useful. It has to be quite specialist, quite bespoke physiotherapy and the chap here from W [Warwick] University today has actually come down to recruit patients to a trial. The overall sort of duration of therapy is roughly around sort of 6 months for both of them to have their benefit, so it’s a similar timescale. Obviously the operation has small risks attached to it. There’s a very small risk of serious nerve injury, I mean it’s less than half a per cent, but, you know that’s the main concern with keyhole hip surgery, because we have to put the leg on traction in order to get in to the hip joint. And sometimes people’s nerves do not like being pulled upon, but that is very rare. We’ve had 2 cases at this hospital in around about 1000, so you know it’s a rare complication. A little bit of wound infection is a possibility, a little bit of bleeding, blood clots, things like this. They are all theoretical risks; they all seem to be quite low for keyhole surgery, so nerve injury is the only sort of serious problem that we’ve experienced. You can get a bit of numbness of the outside of the thigh because of the skin nerve that can get pranged because it’s near one of the portals. So these are the kind of downsides of surgery. So you know those are the two options available and I think you’re a good candidate for either one.

PI8

Graphic descriptions of surgery that may have put patients off randomisation; for example, one clinician explained:

There’s always a risk from the traction that it may stretch the nerves down the leg, so that could leave you with some numbness. If you’re very unlucky it could leave you with a little bit of weakness there.

PI4

Presenting trial information in an order that is confusing for patients, which is illustrated in the following quote:

OK, and I’ll probably bring [surgeon] in to explain more about the pros and cons of each of the treatments. Mr [local surgeon] would do the arthroscopy, but for the purpose of the study, we ask people if they’re happy to do it, and either of the treatments is chosen for you. So a specialised programme, a computer programme will choose the treatment of your . . . that you will receive.

SN004

Surgeons going beyond their protocol brief, to explain the trial rather than referring patients on to the trial recruiter for this information; this is illustrated in the following quote in which the surgeon has already discussed and agreed on a treatment before calling the researcher:

So I’m going to recommend that we (drilling continues) err, at least consider looking inside your hips.

Yes.

If you’ve got a moment, I would appreciate if you have a chat with P, who’s the researcher.

OK.

And just see if the study appeals to you. But you don’t have to, there’s no pressure, you don’t have to be enrolled in the study.

SN008

Conversely, surgeon endorsement of trial participation to patients appeared to have a positive impact on recruitment.

Previous research has shown that, in general, one in four people decline to participate in a RCT despite high-quality recruitment practices because of strong patient preferences or treatment refusal. The recruitment rate for the FASHIoN study was 70%, which suggested processes of informed consent and participant self-selection were working well. Therefore, identifying what worked in the recruitment appointments was relevant in order to replicate recruitment practices in the full trial.

A model of a good FASHIoN recruitment consultation was developed based on communication patterns identified in the observation of successful recruiters and the comparison and contrast of these patterns with the ones observed in unsuccessful recruitment consultations. Aspects such as sequence of presentation, balancing the time and level of detail of the treatment discussions and the sufficiency of explanations about the trial procedures were addressed. Figure 9 shows the stepwise approach to a good a recruitment consultation that has been developed based on the analysis of the audio recordings collected during the FASHIoN trial.

FIGURE 9.

The FASHIoN model of a good recruitment consultation: a six-step approach.

The main principle underpinning the model is that recruitment consultations are different from clinical consultations in that they should enable patients to understand uncertainty and lack of clinical research evidence. In routine clinical consultations, clinicians and patients may be dealing with many uncertainties at the beginning, but these are soon resolved through the interaction and when a decision about treatment options is made. The direction of the clinical consultation is from uncertainty towards decision-making, while in recruitment consultations the sequence of information sharing starts from stating what is currently known about the treatments and moves towards uncertainty. During the observation of successful recruitment consultations, a logical sequence emerged which achieved this purpose and is explained next.

Step 1: explain what femoroacetabular impingement is to the patient

The first step in the recruitment process is the diagnostic consultation. In successful recruitment consultations, patients received an explanation about FAI that was easy to understand. Clinicians tended to use lay terms to illustrate what happens to the body because of this condition (e.g. ‘shape abnormality’, ‘egg shape’, ‘extra little piece of bone’, etc.). They also made use of metaphors from common everyday experiences, for example piston heads in a car, as illustrated by the following quote:

If we imagine a car has a piston that’s not smooth, yours, your hip joint is like a, sort of a joint that’s not completely smooth and so when you try to move it, it’s causing damage to and pain in your hip joint.

SN034

These attempts to make sure the patient understands what it is happening to their bodies are important investments in the relationship and helped patients to feel confident in the care they were receiving. In addition to the ‘shape abnormality’ explanation of FAI, successful recruiters also introduced the role of muscle control in the diagnosis of FAI. This information permitted patients to make a logical link between the condition and the PHT as a plausible treatment, which could be utilised later in the discussion. Many recruiters expressed their concern about not being able to justify why physiotherapy could be effective. This explanation addressed this concern; for example, a surgeon summarised it as:

My idea on this is if you’ve got the egg shape, and your muscles are not good at supporting it, then you run into trouble.

SN053

Surgeons also introduced the common incidence of this condition in the population, which supported the sense of urgency about answering the research questions and increasing the available knowledge about FAI.

Step 2: reassure the patient that they will receive treatment

Patients who agreed to participate in the trial were reassured their diagnosis was confirmed. They were also reassured they would receive the best treatment to meet their individual needs, whether they agreed to be in the trial or not, as illustrated in the following quote:

You’re getting a bit of extra rubbing. And that rubbing is causing the pain, so there’s not really any mystery, we know what the problem is [. . .] my suggestion is that we treat this problem; I don’t think we should just leave it alone, I think we need to try and make you better.

SN034

This was important because many patients waited for a long time to reach the appointment with the surgeon and may have been previously misdiagnosed. Patients may arrive at the clinic dissatisfied with the treatment they received so far. Often patients who declined to be part of the trial have experienced these difficulties. Being told that finally they have the right diagnosis, that someone is confident about it, generates trust and openness to the invitation to help improving care for other patients like themselves. This step addressed the previously identified concern that the treatment allocated to them is not taking into consideration their personal circumstances and their individual needs for care. A direct invitation from the consultant surgeon to the patient about participating in the trial seems particularly effective in getting the person to consider the trial and to listen to the recruiter; an example is presented in the following extract:

And in a minute, one of my colleagues, if you agree, will come and talk to you about how that particular study is working. And, if you would like to, I would very much like to include you in that study?

SN053

Step 3: explain that there is uncertainty about which treatment is the best

Steps 1 and 2 promote certainty, whereas step 3 is about introducing uncertainty. In successful recruitment consultations, uncertainty was mentioned earlier on and by the consultant, and then it was reinforced repeatedly by the recruiter during their appointment with the patient. Even in effective recruitment consultations, recognising uncertainty may be difficult for patients. Effective recruiters enabled the patient to understand that the two treatments that were being compared are effective in their own right. This argument helps to compensate for the uneasiness of not knowing which treatment is best. The following extract of a diagnostic consultation exemplifies this step:

I think we can offer you some treatment for it, but the trouble is we’ve only really just started to understand this disease [. . .] we’ve got two treatments that we know can help, but we just don’t know which is the better one.

What we’re doing is, as [the consultant] has explained, we don’t know which is best, we’re not sure. [Consultant], myself and any of the other consultants that specialise in hips here, can’t tell you. For you, I think the best thing to do is X or Y because we don’t know.

You don’t know what the best treatment is?

Well, we need people like you to say, ‘OK, I’m happy to take part, and whichever treatment I get then I should, should get a little bit better’, but one will be better than the other probably.

SN034

Step 4: explain the purpose of the study

Once uncertainty was been explained, the purpose of the study follows logically. Recruiters stated how the findings of the research would help clinicians to advise patients like them in the future. At this point, successful recruiters mentioned the need for evidence, implying the real contribution to advancing science and health care that patients could make. Recruiters were successful at harmonising the message that FAI is a condition that clinicians are interested in knowing more about how to treat and the message that patients would be valued, respected and cared for during the trial. An example is presented below:

[The recruiter] can explain the study to you, and explain how we try and find out the answer to the question and how you might get involved in that. One thing though, is whatever we do, and I’m going to do everything I can to make you better.

SN006

Step 5: give the patient a balanced view about the pros and cons of each of the two treatments

In some cases, the recruitment consultation was introduced as a conversation about the options available to patients. For example, one surgeon said:

[Recruiter] is doing a study looking at the different solutions and he could talk to you about the options if we got to the point where you said you’d like to do something about it.

SN007

Instead, some successful recruiters and clinicians avoided the word ‘options’ when referring to the treatment arms of the RCT because it implies there is a decision to be made between the two treatments, whereas the patient has to decide between being part of the trial or not. Alternative expressions such as ‘available treatments’ or ‘ways to treat FAI’ were used, providing a clearer message. In effective recruitment consultations, the treatment arms were presented as two ways of dealing with the same condition. Recruiters spent similar amounts of time explaining each one of the arms. They tended to start with the non-operative arm and the balance between the two treatments was maintained when talking in detail about the benefits and risks of each arm (e.g. number of visits to hospital for each arm, duration of the intervention). For example, a recruiter discussed the balance between arms as:

The kind of use of your time is equal, whichever one you go for, whichever one you’re allocated to . . . The reason I say that is because if you have an operation you have to come for a pre-operative assessment, you have to have the operation and you have to have some rehab. If you’re allocated to the personalised hip therapy, you’ll have to come here on three, at least three occasions, and then they’ll give you specific exercises to practise at home.

SN034

Effective recruiters also emphasised the benefits of the two treatments in order to assure patients they were not taking any more risks than if they were making the choice themselves. After step 5, patients usually were not sure what treatment to choose. They may have thought about having one treatment at the start of the conversation, but at this point, they expressed either the opposite choice or simply uncertainty about which one to have. Patient equipoise was thus achieved.

Step 6: explain the study procedures

Once uncertainty and equipoise were established, the allocation of treatments by randomisation seemed more acceptable. Randomisation was mentioned later on in the conversation and recruiters spent a lot less time explaining it than the time they spent on the other steps. Randomisation was usually presented as an allocation or being invited to join one of the groups, which should allow a fair comparison. The following quote exemplified this explanation:

But in order to make it a fair test and in order to make sure that we’re truly, truly testing which one’s better, we have to go through a process whereby we allocate you with a treatment. So I don’t choose and [Professor] doesn’t choose and we don’t choose which, which treatment you have . . . it’s randomly allocated to you

SN003

The questionnaires and the reassessment after a year were constructed as part of a closer follow-up of participants in the trial than for people going through regular care. This was part of the efforts of the recruiters to emphasise that patients would be treated with respect and receive personalised care.

Two more tasks were added to the model. They are not a fixed part of the six-step sequence; instead they happen at different points throughout the most successful consultations. These two tasks appeared to be essential and determinant in the successful of a recruitment consultation.

Task 1: respond to patients’ concerns and questions

The majority of patient questions occurred after the two treatments were explained. Often patients wanted to know about the details of each arm and the rationale for the non-operative arm. Successful recruiters stopped delivering information and answered the questions or listened to patient concerns before moving onto the next step. The conversations in the most successful recruitment consultations were highly patient-centred, with evidence the clinicians were responding to the patient’s concerns and doubts about the trial. Patients asked more questions than in unsuccessful recruitment appointments. Recruiters confirmed that patient participation was voluntary and that patients would receive treatment regardless of their involvement with the study, which encouraged trust.

Task 2: show confidence and a relaxed manner

Effective recruiters appeared confident and relaxed when talking about the study. This was based on having a good working knowledge about the condition and the treatments or taking actions towards achieving it. In addition, successful recruiters used a set of ready-made answers to common patient questions prepared by the TMG and worked closely with the clinicians in order to deal with tricky questions.

The model structure is identical to the approach used in the PIS. Therefore, giving the patient the opportunity to read this document at the beginning of recruitment could save time dedicated to information sharing. Instead, this time can be used to respond to patients’ concerns, showing the balance between risks and benefits between the two trial arms, reassuring patients about participating in a trial, explaining the condition in more detail or discussing uncertainty. Feedback to the sites was prepared based on this model and presented to the recruiters and PIs (see Evidence base for the trial). Further data collection continued to be used to develop and validate the model.

Method

Semistructured interviews were conducted with a sample of patients who have received treatment for FAI. The interviews explored patients’ perspectives on their hip problems, previous experiences with treatments, views about arthroscopy surgery and conservative care and the acceptability of randomisation between the interventions. Fourteen patients interviewed were also asked to complete the expert patient survey while talking to the researcher.

Findings

Findings

Recommendations

Clinicians’ interviews

The purpose of this activity was to explore attitudes to randomisation of FAI patients of a multidisciplinary sample of clinicians working in the centres performing hip arthroscopy. Semistructured interviews were carried out and, to stimulate discussion, clinicians were provided with anonymous patient cases that cover the spectrum of patient presentations, including patient history (with duration of symptoms and previous treatments), examination findings and imaging (for examples and interview schedule please see Appendix 1). By introducing patient cases we sought to deconstruct the cognitive processes involved when a clinician considers recruiting a patient to a clinical trial. This was done in order to elucidate the state of individual equipoise that may be influencing a recruitment decision.

Findings

Recommendations

1. Clinicians asked for clear inclusion and exclusion criteria. These should include requirements that can be interpreted by a multidisciplinary team. Surgeons, physiotherapists and sport doctors tend to work together.

2. Surgeons are concerned about accurate classification of morphology and degenerative changes in order to exclude similar conditions. Physiotherapists would like more information about neuromuscular problems and other examination parameters.

3. Reaching agreements about the age range, level of activity, severity of symptoms and even psychological assessment of patients will be useful, as they are important cues in deciding what treatment to suggest.

4. Concerns about NHS patients who usually have tried non-operative care plans and may ask for the surgical treatment need to be addressed. The non-operative arm has to be presented as a different option from past care plans, to the clinicians as well as to the patients.

5. Information about outcome measures and how the results will be disseminated can be a motivational tool.

6. Clarification about how the fact that patients have had past therapy will affect the results of the trial may be necessary.

7. In the process of choosing the centres where the trial will be run, and the surgeons who will take part, there has to be awareness about possible vested interests in showing that the surgery works better, willingness to participate in research and quality of the local physiotherapy.

8. Addressing the concern of deterioration during the 12-month period would be helpful, for instance by explaining the procedure of regular and continuous assessment of patients’ condition and also showing that patients will be informed about the risks of delaying treatment.

9. It may be advisable to provide more information about how the non-operative arm will be presented to patients and how patients’ expectations will be managed to minimise drop-out rates owing to lack of results.

Background

• How long have you recruited patients to RCTs?

• How did you become a recruiter?

• Have you recruited to other RCTs, before this one?

• How did you come to be working on this RCT?

• Have you received any training (a) as a recruiter; (b) for this RCT?

• When did you start recruiting to this trial?

• Please describe how recruitment has been for you in this RCT.

Introducing/explaining the trial

• How do you introduce/explain the trial to patients?

• Why do you think this RCT is needed?

• What is involved in taking part in this RCT; and in each arm?

• What happens to patients if they agree to take part?

Randomisation

• How do you explain to patients how the decision is made in the trial about which treatment they will receive?

• How easy do you think the concept of randomisation is for patients to understand? Note: If this seems threatening, acknowledge that this is a difficult and confusing concept for everybody and then explore again.

• Does randomisation seem sensible and reasonable to you?

Uncertainty

• How do patients react to the idea that the specialist doctor does not know what the best treatment is?

• How do you describe the uncertainty?

• How do you feel about explaining uncertainty?

• Do you ever have a feeling during an appointment that a patient should really have one treatment rather than another? Probe: Why? What do you do about that?

Difficulties during recruitment

• What would you say are the main difficulties you face as a recruiter?

• Can you describe specific examples of ‘good’ and ‘bad’ recruitment experiences?

• Is recruitment organised well? Do other people explain the RCT to patients?

• Are there difficulties with any particular arm?

• Do patients have strong preferences?

• Do you think this RCT is the right thing for these patients?

Finally – improving recruitment

• What might, in your opinion, improve the recruitment appointments?

• Do you think you (or others) need more support or training?

Recruitment pathway

• How are patients recruited into this RCT? Is there a ‘recruitment’ appointment?

• Who undertakes the assessment of patient eligibility for this RCT? Are you involved in eligibility assessment?

• How do you find out which patients to approach about the RCT?

Informed consent

• How easy do you think it is to obtain fully informed consent for patients participating in this RCT? (Probe: what is informed consent; how do you know when you have reached it?)

• Are you ever concerned about the possibility of coercing patients into taking part in the trial?

Treatment preferences of patients

• Do patients express preferences for particular treatments?

• (Probe – what are their preferences; do you know why they express preferences?)

• What do you do when a patient expresses a preference?

• (Probe – do you accept it – what happens next? Or do you ever explore or challenge their preference?)

• What other reasons (not preferences) have patients given for not taking part in this RCT?

• What do you do when a patient gives a reason for not wanting to take part?

Personal views

• Do you think your work as a recruiter is similar or different from your usual work/practice? Probe: In what ways it is (a) similar, (b) different?

• Do you think your work as a recruiter has changed the way you practise outside research?

• How do you feel about being a recruiter? Probe: Do you see yourself as a doctor/nurse/clinician, scientist, or researcher (if nurse also add ‘patient advocate’); or all of these? Are there benefits or problems that arise from these different roles?

• Do you know (or have a hunch) about what the outcome of the RCT will be?

• If you were a patient in this position today, would you agree to be recruited to the RCT and be randomised? Or would you choose a treatment? Which one?